NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/MiniMax M2.5
minimax logo

MiniMax M2.5

by minimax · Released Feb 2026

Open Source
51.4
avg score
Rank #135
Compare
Better than 57% of all models
Context
205K tokens (~102 books)
Input $/1M
$0.27
Output $/1M
$1.08
Type
text
License
Open Source
Benchmarks
24 tested
Data as of
About

MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digital working environments, M2.5 builds upon the coding expertise of M2.1...

Tested on 24 benchmarks · BenchGecko score 51.4. Top scores: Chatbot Arena Elo — Overall (1390.8%), Chatbot Arena Elo — Coding (1386.3%), OpenCompass — IFEval (91.1%).

Looking for similar performance at lower cost?
Gemma 2 9B scores 51.1 (99% as good) at $0.03/1M input · 89% cheaper
Capabilities
coding
59.7
#49 globally
reasoning
44.4
#91 globally
math
81.8
#22 globally
knowledge
51.6
#113 globally
agentic
6.2
#62 globally
language
67.8
#87 globally
general
7.2
#202 globally
Benchmark Scores
Compare All
Tested on 24 benchmarks · Ranked across 8 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
OpenCompass — LiveCodeBenchV6

OpenCompass Live Code Bench v6. Fresh competitive programming problems to evaluate code generation without memorization.

73.6·
LiveBench — Coding

Regularly refreshed coding problems that avoid data contamination. New problems added monthly to prevent memorization.

70.7·
LiveBench — Agentic Coding

LiveBench coding tasks that require multi-step reasoning and tool use. Tests planning and execution of complex coding workflows.

51.7·
reasoningCompare reasoning →
ARC-AGI

Abstraction and Reasoning Corpus. Tests fluid intelligence through novel visual pattern recognition puzzles. Core measure of general intelligence.

63.7·
LiveBench — Reasoning

Regularly refreshed reasoning problems testing logical deduction, spatial reasoning, and analytical thinking.

59.3·
LiveBench — Data Analysis

Fresh data analysis tasks testing ability to interpret tables, charts, and statistical data.

49.6·
mathCompare math →
OpenCompass — AIME2025

OpenCompass evaluation on AIME 2025 problems. Tests mathematical reasoning on fresh competition problems.

86.2·
LiveBench — Mathematics

Regularly updated math problems that test numerical reasoning, algebra, calculus, and combinatorics.

77.4·
Quick compare:OpenAI logovs o4 MiniDeepSeek logovs DeepSeek V3.2minimax logovs MiniMax M2.7
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · minimax MiniMax M2.5
Feb 2026
51.4
$0.27/M in205Kctx24 benchmarks
Feb 2026
$0.00/M in(-0.27)197Kctx(-8K)
See the full MiniMax M2.5 family →
Recently Happened
MiniMax M2.5 pricing dropped 10%
Sep 13, 2026
MiniMax M2.5 pricing increased 11%
Sep 10, 2026
MiniMax M2.5 pricing increased 14%
Aug 22, 2026
MiniMax M2.5 pricing increased 88%
Jul 10, 2026
MiniMax M2.5 pricing dropped 47%
Jun 26, 2026
MiniMax M2.5 pricing increased 27%
Apr 21, 2026
Similar Models
OpenAI logo
o4 Mini
OpenAI
51.4$1.10/1M
DeepSeek logo
DeepSeek V3.2
DeepSeek
51.3$0.28/1M
minimax logo
MiniMax M2.7
minimax
51.2$0.21/1M
Links
Info
minimaxPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@minimax
Source Code
GitHubHugging Face
BenchGecko API
minimax-m2-5
Specifications
  • Typetext
  • Context205K tokens (~102 books)
  • ReleasedFeb 2026
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.002
Available On
minimax logominimax$0.27
Categories
codingreasoningmathknowledgeagenticlanguagegeneral
Learn More
context-windowtransformeropen weightstokens
Share & Export
Tweet
Related Models
OpenAI logoo4 Mini
51.4
DeepSeek logoDeepSeek V3.2
51.3
minimax logoMiniMax M2.7
51.2
DeepSeek logoDeepSeek V3.2 Exp
51.6
MiniMax M2.5 is an open-source text AI model by minimax, released in February 2026. It has an average benchmark score of 51.4. Context window: 205K tokens.

Related Models

o4 Mini · OpenAIDeepSeek V3.2 · DeepSeekMiniMax M2.7 · minimaxDeepSeek V3.2 Exp · DeepSeekGemma 2 9B · Google DeepMind

Benchmarks

Chatbot Arena Elo — OverallChatbot Arena Elo — CodingOpenCompass — IFEvalOpenCompass — AIME2025OpenCompass — GPQA-Diamond

Related Pages

minimax · Providerminimax · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-10-05

  • MiniMax M2.5 by minimax. BenchGecko score 51.4, rank 134 of 312 scored models (normalized average of public benchmark scores).
  • List price $0.27 input · $1.08 output per 1M tokens (as of 2026-10-05).
  • Sold by 9 providers (as of 2026-10-05): StreamLake $0.27 in / $1.08 out · Venice $0.27 in / $0.95 out · AtlasCloud (fp8) $0.29 in / $1.20 out · DigitalOcean $0.30 in / $1.20 out · Friendli $0.30 in / $1.20 out · and 4 more. Every provider
  • Gecko Tests: Who Are You E (Sometimes says it is OpenAI) · Tokenizer Tax D (64% more tokens outside English).

How to cite · data as of 2026-10-05

MiniMax M2.5 · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-05. https://benchgecko.ai/model/minimax-m2-5

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP